As far as I understand, a number of things make a difference:
- Socketed RAM is limited by the number of pins, DDR SO-DIMM for example has 2-3x fewer pins than your average LPDDR4X module. Having more pins helps running the chips on lower power as it was explained to me by an engineer (something to do with grounding and lower voltage, I am not really sure how it works but I trust the guy)
- Socketed RAM is a least common denominator, there will inevitably be some variance between different modules, which leaves you with less possibilities for optimizing the electrical connection
- Current LPDDR is much more sophisticated than regular DDR, the protocol is different etc. So you can't just stick LPDDR chips on a DDR SO-DIMM and expect it to work
I am sure one could design a modular system based on LPDDR, but the question is whether such system would be feasible. It would be limited to premium laptops only (as modular system would take too much space in a smartphone), it would probably need a higher pinout (which would mean a complex mounting bracket of some sort) and it would require laptop manufacturers to agree on a certain standard.
There are other considerations as well, especially if one wants to break the current performance boundaries. For example, Intel has announced that they will integrate HBM with their upcoming Xeons. How do you intend to make that socketable, that's 2048-bit memory bus. Or consider Apple's new in-house SoCs. By mounting RAM directly onto the package substrate they can potentially deliver high-bandwidth RAM in an energy efficient package without increasing the cost and the complexity of the mainboard. E.g. their upcoming hardware is widely expected to use 256-bit RAM, something that has so far been reserved for workstations. As you make the interface wider and wider, modularity becomes more and more expensive. Why don't GPU's offer modular RAM for example?
Modern CPUs are an order of magnitude more complex and usually built on a smaller process, yet we have no trouble putting them on intermediary PCBs to spread out contacts and allow larger pins. I see no reason we couldn't make LGA sockets for DRAM chips in the same way.
The beauty and success of DIMM modules is that they are a simple slide-in to install technology. It's easy to operate, cheap, robust, and it is very space efficient. Imagine RAM modules in an LGA socket instead... how much space would it require? How robust would it be? What will be the price? I am quite sure all of these issues can be solved, but is it worth it? Let's be honest, users who are interested in having upgradeable RAM in their ultraportable laptop are such a small minority that the practical interest in developing modular low-powered RAM is close to zero.
The practical interest might be close to zero because in the past memory upgrades a number of years down the road were one of the most common upgrades and added tremendous value and another few years of life. Similarly many older laptops got a second life by being upgraded to SSDs. Accidents like that, which prevent sales of new units, can't happen if storage and memory are soldered on.
I don't think this is the driver of these decisions per se, but it is undeniably a bonus for the manufacturer; non-upgradable devices become obsolete faster, necessitating new replacements.
I don't know about other brands, but for ThinkPads (both in the IBM and Lenovo eras), there was often a stated maximum RAM that was based only on the capacity of SODIMMs available at the time of manufacture.
But it wasn't an artificial limitation. In practice, once larger capacity memory became available with the same technology and form factor, it would work fine. I have upgraded several ThinkPads with memory beyond what the original datasheets said was possible.
I have absolutely no idea, but I guess the bus to the processor and other periphery could indeed be a weak point because of EM-interference. We have absolutely crazy data rates here and therefore very high frequency signals so proximity between components might indeed matter.
That said, manufacturers have huge ambition to not allow for modular design.
- Socketed RAM is limited by the number of pins, DDR SO-DIMM for example has 2-3x fewer pins than your average LPDDR4X module. Having more pins helps running the chips on lower power as it was explained to me by an engineer (something to do with grounding and lower voltage, I am not really sure how it works but I trust the guy)
- Socketed RAM is a least common denominator, there will inevitably be some variance between different modules, which leaves you with less possibilities for optimizing the electrical connection
- Current LPDDR is much more sophisticated than regular DDR, the protocol is different etc. So you can't just stick LPDDR chips on a DDR SO-DIMM and expect it to work
I am sure one could design a modular system based on LPDDR, but the question is whether such system would be feasible. It would be limited to premium laptops only (as modular system would take too much space in a smartphone), it would probably need a higher pinout (which would mean a complex mounting bracket of some sort) and it would require laptop manufacturers to agree on a certain standard.
There are other considerations as well, especially if one wants to break the current performance boundaries. For example, Intel has announced that they will integrate HBM with their upcoming Xeons. How do you intend to make that socketable, that's 2048-bit memory bus. Or consider Apple's new in-house SoCs. By mounting RAM directly onto the package substrate they can potentially deliver high-bandwidth RAM in an energy efficient package without increasing the cost and the complexity of the mainboard. E.g. their upcoming hardware is widely expected to use 256-bit RAM, something that has so far been reserved for workstations. As you make the interface wider and wider, modularity becomes more and more expensive. Why don't GPU's offer modular RAM for example?